Skip to main content
Yash Djson Dookun
Based in Mauritius ยท Building for SMEs

Custom business systems, APIs, and AI-powered tools for SMEs.

I help small and medium businesses reduce manual work, centralize operations, and unlock practical AI โ€” without enterprise budgets or vendor lock-in.

Senior IT Consultant ยท 5+ years experience ยท BSc & MSc from University of Mauritius

Technologies I work with

Python
FastAPI
Streamlit
PostgreSQL
SQLite
REST APIs
Docker
LLMs / RAG
CLI Tools
Data Pipelines

What I Build

internal-tools

Internal Business Tools

Custom tools for quotations, inventory, booking, reporting, document tracking, and daily operations.

backend-apis

Backend APIs

REST APIs and backend services that can be consumed by PowerApps, Power Automate, Excel, dashboards, or other internal tools.

data-reporting

Data & Reporting Systems

Clean messy data, structure CSV/Excel files, build reports, and create dashboards for operational visibility.

websites

Lightweight Business Websites

Simple professional websites for credibility and contactability. Clean, fast, and ready to go live.

knowledge-search

Knowledge & Search Systems

Document search, RAG-style retrieval, internal knowledge bases, and searchable repositories.

monitoring

Monitoring Tools

Small systems that track public/internal data sources and expose structured information.

Services

Scoped, practical, no bloat.

01

End-to-End Business Systems

Full custom systems scoped to your operations โ€” from database to interface. Built for SMEs that need something purpose-built, not another subscription tool that almost fits.

Inventory management systems Quotation & invoicing tools Booking & scheduling platforms Customer management systems Document tracking workflows Operational dashboards

Scoped to what your business actually needs. No bloated features you won't use โ€” just the system that solves the problem.

02

Technical Consulting & Architecture

Strategic guidance on technology decisions, system design, and infrastructure planning. Ideal for businesses that need expert advice before committing to a build.

System architecture reviews Technology stack selection Migration planning Infrastructure recommendations Technical due diligence Performance audits

Advisory and planning work. Implementation is scoped and quoted separately if you decide to move forward.

03

Backend APIs & Integration Layers

Backend services and APIs that connect your tools, automate handoffs, and expose data where it's needed.

REST APIs Database-backed services CSV/Excel import & export API documentation Webhook-ready endpoints Integration with client-managed tools

I build the backend contract. Clients remain responsible for their own PowerApps, Power Automate, Excel, or low-code environment.

04

Data Cleanup & Reporting Tools

Turn messy CSV/Excel data into clean, structured, usable datasets and dashboards. Give your business visibility into what's actually happening.

Deduplication Standardization Reporting exports Operational dashboards Internal analytics Lead scoring pipelines

I structure and clean the data. Ongoing data entry and maintenance stays with your team.

05

AI-Powered Tools & Automation

Practical AI that saves your team hours โ€” not hype, but real tools that use LLMs and generative AI to automate document processing, search internal knowledge, or assist with repetitive tasks.

Document classification & extraction Internal knowledge search (RAG) AI-assisted data entry Automated report generation Intelligent workflow routing Custom AI agents for business tasks

Built around your real workflows. I scope what AI can and can't solve for your case before building anything.

06

Lightweight Business Websites

Simple, professional websites that give your business a credible online presence and make it easy for clients to reach you.

Static website Contact links WhatsApp button Business/service information Hosting setup

Focused on design, structure, and hosting โ€” marketing and ad campaigns are managed separately by the client.

How I Work

From first conversation to delivered system.

01

Describe the workflow

You tell me what process is painful โ€” manual data entry, scattered spreadsheets, a missing dashboard, or a system that doesn't exist yet.

02

I scope the system

I define what gets built, the deliverables, the stack, and the timeline. You get a clear, written proposal before anything starts.

03

Build & iterate

I build in short cycles with regular check-ins. You see progress, give feedback, and the system takes shape around your real needs.

04

Deliver with docs

You get the finished system, documentation, and everything needed to run and maintain it independently.

5

Systems built

11K+

Records processed

3

Data pipelines

855

Documents indexed

Scope & Focus

To deliver well, I keep a clear focus. Here's what falls inside and outside my work.

Custom tools to replace manual processes

Not: Social media management

Dashboards & reporting for your operations

Not: Facebook or Google ads

AI tools that save your team time

Not: Branding or graphic design

Data cleanup, structuring & pipelines

Not: SEO campaigns or ad management

APIs & system integrations

Not: RPA bot development

Technical architecture & consulting

Not: PowerApps / Power Platform maintenance

I build the systems, tools, and automation that help your business run better โ€” but the goal is reliable software, not marketing management or low-code platform maintenance.

Selected Systems & Technical Work

Real systems built for real problems โ€” data pipelines, monitoring tools, legal tech, and AI automation.

Problems I solve

  • Replace spreadsheet-heavy workflows
  • Centralize fragmented processes
  • Build internal business tools
  • Expose clean APIs
  • Turn messy data into usable information
KotMoKouran
$ python -m app fetch ceb --district plaines-wilhems
Fetching CEB outage data...
District: Plaines Wilhems
โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”
โ”‚ ACTIVE OUTAGE โ”‚
โ”‚ Locality: Rose Hill โ”‚
โ”‚ Streets: Rue St-Jean, Ave Berthaud โ”‚
โ”‚ Period: 08:30 โ€” 15:00 โ”‚
โ”‚ Status: active โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜
3 active | 2 upcoming | 5 cleared today

KotMoKouran

Power Outage Monitoring System

A monitoring system that collects CEB outage data for Mauritius, normalizes it into a structured format, and makes it easy to query by district, locality, and status. Built for teams that need reliable visibility into planned and active outages โ€” not scattered social posts or manual checks.

What this delivers

  • Real-time outage tracking
  • Public utility data aggregation
  • Monitoring and reporting
  • Built specifically for Mauritius
CEB Site
Scraper
Parser
Normalizer
Structured Data
MonitoringPublic DataMauritiusPython
Python63% TypeScript28% Shell4% PowerShell3% JavaScript1%
View on GitHub
ClientList
$ python -m clientlist pipeline --score --export
Source: SME e-Directory (Mauritius)
Scraped: 11,247 raw records
Normalized: 11,103 cleaned
Deduplicated: 9,831 unique businesses
Scored: 2,143 high-priority leads
Validated: 1,876 with working websites
Exported: clients_scored.csv (9,831 rows)
Exported: high_priority.json (2,143 leads)

ClientList

Mauritius SME Intelligence Platform

An end-to-end data pipeline that turns the national SME e-Directory into outreach-ready business intelligence. Scrapes raw records, cleans and deduplicates them, validates websites, scores leads by priority, and exports structured CSV/JSON for sales, research, or internal CRM import.

What this delivers

  • 11,000+ businesses processed
  • Data cleaning & deduplication
  • Lead scoring & segmentation
  • Website validation
  • CSV & JSON exports
SME Directory
Scraper
Normalizer
Dedup
Scorer
CSV/JSON
PythonData EngineeringCLILead Scoring
Python96% HTML2% Shell2% Makefile0%
View on GitHub
MorisLex-Engine
$ python -m app.services.engine_main --run full
Sources:
Laws of Mauritius โœ“ 412 documents
SAFLII Mauritius โœ“ 287 judgments
National Assembly โœ“ 156 bills
Pipeline: fetch โ†’ extract โ†’ classify โ†’ dedup
New: 23 | Updated: 7 | Unchanged: 825
Total corpus: 855 documents

MorisLex Engine

Mauritius Legal Knowledge Platform

A configurable legal data platform that ingests Mauritian legislation, judgments, and parliamentary content from multiple public sources. Uses adapter-based scraping, classification, and deduplication to maintain a structured corpus โ€” with a Streamlit interface for operations and exports ready for search or RAG downstream.

What this delivers

  • Built searchable legal corpus
  • Automated legal data ingestion
  • Configurable scraping engine
  • CSV/YAML-driven configuration
  • Streamlit management interface
  • RAG-ready outputs
Legal Sources
Adapters
Extractor
Classifier
Dedup
Corpus
PythonScrapingLegal DataStreamlit
Python95% Makefile2% Shell2% Jinja1% Dockerfile0%
View on GitHub
MorisLex-Rag
$ python -m app query "data protection employer obligations"
Retrieving from 855 indexed documents...
Model: all-MiniLM-L6-v2 | Top 3 results
[0.94] Data Protection Act 2017 โ€” Section 28
"An employer processing personal data shall..."
[0.87] DPA 2017 โ€” Section 36(1)
"Data controllers must implement appropriate..."
[0.81] Employment Rights Act โ€” Section 13A
"No employer shall require an employee to..."

MorisLex RAG

Semantic Legal Search & RAG System

A retrieval-augmented generation layer on top of the MorisLex legal corpus. Chunks and embeds documents, indexes them for semantic search, and answers natural-language questions with cited sources โ€” useful for compliance research, internal legal Q&A, or policy teams that need fast answers without reading hundreds of pages.

What this delivers

  • Semantic legal search
  • Retrieval-Augmented Generation (RAG)
  • Document indexing & embeddings
  • Natural language querying
  • Private knowledge base architecture
  • AI-assisted legal retrieval
Engine Corpus
Chunker
Embedder
Vector Store
Retriever
LLM
RAGLLMSearchEmbeddingsLegal Tech
Python82% Shell15% Makefile3% Dockerfile1%
View on GitHub
Agentic-Poc
$ python -m agentic run --workflow document-review
Workflow: Document Review Automation
Agents: 3 (classifier, extractor, validator)
Step 1/4 Classify incoming documents โœ“
Step 2/4 Extract structured fields โœ“
Step 3/4 Validate against schema โœ“
Step 4/4 Route to review queue โœ“
Processed: 48 documents in 12.3s
Accuracy: 94.2% (validated against manual)

Agentic-Poc

Agentic AI Initiative

An agentic AI proof of concept pitched internally and adopted at center level. Designs multi-step workflows where specialized agents classify, extract, and validate documents โ€” demonstrating how SMEs and enterprise teams can move from manual review to orchestrated AI automation with measurable accuracy.

What this delivers

  • Concept pitched internally
  • Adopted at center level
  • AI enablement focus
  • Cross-team collaboration
  • Multi-agent workflow orchestration
  • Document review automation
Workflow
Agent Config
Orchestrator
Validation
Output
Agentic AIPoCAI EnablementSolution Design
Python100% JavaScript0%
View on GitHub
About

Yash Djson Dookun

Senior IT Consultant ยท Mauritius

I'm a Senior IT Consultant with 5+ years of experience across enterprise consulting and software development. I hold a BSc in Web & Multimedia Development (First Class Honours) and an MSc in Network Security & Management from the University of Mauritius.

My day-to-day involves designing, building, and integrating custom software solutions across multiple system components. Outside of that, I build practical systems for SMEs in Mauritius โ€” the kind of internal tools, APIs, data pipelines, and AI-powered automation that help small businesses operate more efficiently without enterprise budgets.

I work independently on these projects, which means direct communication, no handoffs, and full ownership from scoping to delivery.

Working Principles

1

Clear scope before building

Define deliverables and boundaries upfront.

2

Simple architecture over complexity

No over-engineering. Solve the actual problem.

3

Maintainable systems

Code someone else can understand and extend.

4

Documentation included

Every system ships with docs, not just code.

5

Built for low-maintenance operation

Systems that run without constant attention.

I prefer small, well-scoped systems that solve specific operational problems. The goal is not to build bloated enterprise software, but to create tools that businesses can actually use and maintain.

Common Questions

Honest answers. No sales pitch.

How much does a project typically cost?
It depends on scope, but most projects for SMEs fall between Rs 15,000โ€“80,000. A simple website or data cleanup tool is on the lower end. A full inventory system or AI-powered tool is on the higher end. I always scope and quote before any work starts โ€” no surprises.
How long does it take?
Most projects take 2โ€“6 weeks. A basic website might be done in under a week. A full business system with multiple features usually takes 3โ€“5 weeks. I work in short cycles so you see progress throughout, not just at the end.
I'm not technical โ€” is that a problem?
Not at all. Most of my clients aren't. You describe the problem in your own words โ€” "we track everything in a spreadsheet and it's a mess" โ€” and I translate that into a system. You don't need to know anything about APIs or databases.
Do I need to prepare anything before we start?
Just a clear picture of the problem you want solved. If you have sample spreadsheets, documents, or screenshots of your current process, that helps. But I can also help you figure out the scope during our first conversation.
Can you work with tools we already use?
Yes. If you use Excel, Google Sheets, WhatsApp, or any tool with an API, I can usually build something that connects to it. I won't ask you to throw away what's working โ€” I build around it.
What happens after delivery?
Every project ships with documentation and a handover. If something breaks in the first month, I fix it at no extra cost. Beyond that, I offer ongoing support on a case-by-case basis โ€” but the goal is to build systems that run without needing me.
Why not just use an off-the-shelf tool?
If one fits, you should. But if you've tried Notion, Trello, Zoho, or Airtable and still end up back in spreadsheets, it's probably because your workflow doesn't fit a generic tool. That's exactly what I build for.
What if the scope changes midway?
It happens. Small adjustments are normal and expected. If the scope grows significantly, I'll flag it, re-scope, and quote the additional work before proceeding. No one gets blindsided.

Get in Touch

Have a manual workflow, spreadsheet, or internal process that is becoming difficult to manage?